Nonparametric Estimation of Density and Hazard Rate Functions when Samples are Censored.
Abstract
The purpose of this article is to present the different types of nonparametric density estimates that have been proposed for the situation that the sample data are censored or incomplete. This type of data arises in many life testing situations and is common in survival analysis problems. Many of the methods of nonparametric density and hazard rate estimation from right-censored observations are discussed. These include histogram and kernel-type procedures, likelihood methods, Fourier series methods, and Bayesian nonparametric approaches. Examples of kernel density estimates are given for mechanical switch life data where data-based choices of the bandwidth values are used. Originator-supplied keywords included: Nonparametric density estimation; Random censorship; Failure rate; Kernel density estimator; Likelihood methods.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 11, 1985
- Accession Number
- ADA150946
Entities
People
- William J. Padgett
Organizations
- University of South Carolina